From: Peter Zijlstra <peterz@infradead.org>
To: x86@kernel.org
Cc: linux-kernel@vger.kernel.org, djwong@kernel.org,
yujie.liu@intel.com, tglx@linutronix.de, jpoimboe@kernel.org,
joao.moreira@intel.com, samitolvanen@google.com
Subject: [PATCH 6/5] objtool: Optimize elf_dirty_reloc_sym()
Date: Wed, 2 Nov 2022 22:46:44 +0100 [thread overview]
Message-ID: <Y2LlRA7x+8UsE1xf@hirez.programming.kicks-ass.net> (raw)
In-Reply-To: <20221028194022.388521751@infradead.org>
Subject: objtool: Optimize elf_dirty_reloc_sym()
From: Peter Zijlstra <peterz@infradead.org>
Date: Wed Nov 2 22:31:19 CET 2022
When moving a symbol in the symtab its index changes and any reloc
referring that symtol-table-index will need to be rewritten too.
In order to facilitate this, objtool simply marks the whole reloc
section 'changed' which will cause the whole section to be
re-generated.
However, finding the relocs that use any given symbol is implemented
rather crudely -- a fully iteration of all sections and their relocs.
Given that some builds have over 20k sections (kallsyms etc..)
iterating all that for *each* symbol moved takes a bit of time.
Instead have each symbol keep a list of relocs that reference it.
This *vastly* improves build times for certain configs.
Reported-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
---
tools/objtool/elf.c | 27 ++++++++++-----------------
tools/objtool/include/objtool/elf.h | 2 ++
2 files changed, 12 insertions(+), 17 deletions(-)
--- a/tools/objtool/elf.c
+++ b/tools/objtool/elf.c
@@ -356,6 +356,7 @@ static void elf_add_symbol(struct elf *e
struct rb_node *pnode;
struct symbol *iter;
+ INIT_LIST_HEAD(&sym->reloc_list);
INIT_LIST_HEAD(&sym->pv_target);
sym->alias = sym;
@@ -557,6 +558,7 @@ int elf_add_reloc(struct elf *elf, struc
reloc->sym = sym;
reloc->addend = addend;
+ list_add_tail(&reloc->sym_reloc_entry, &sym->reloc_list);
list_add_tail(&reloc->list, &sec->reloc->reloc_list);
elf_hash_add(reloc, &reloc->hash, reloc_hash(reloc));
@@ -573,21 +575,10 @@ int elf_add_reloc(struct elf *elf, struc
*/
static void elf_dirty_reloc_sym(struct elf *elf, struct symbol *sym)
{
- struct section *sec;
-
- list_for_each_entry(sec, &elf->sections, list) {
- struct reloc *reloc;
-
- if (sec->changed)
- continue;
+ struct reloc *reloc;
- list_for_each_entry(reloc, &sec->reloc_list, list) {
- if (reloc->sym == sym) {
- sec->changed = true;
- break;
- }
- }
- }
+ list_for_each_entry(reloc, &sym->reloc_list, sym_reloc_entry)
+ reloc->sec->changed = true;
}
/*
@@ -902,11 +893,12 @@ static int read_rela_reloc(struct sectio
static int read_relocs(struct elf *elf)
{
+ unsigned long nr_reloc, max_reloc = 0, tot_reloc = 0;
struct section *sec;
struct reloc *reloc;
- int i;
unsigned int symndx;
- unsigned long nr_reloc, max_reloc = 0, tot_reloc = 0;
+ struct symbol *sym;
+ int i;
if (!elf_alloc_hash(reloc, elf->text_size / 16))
return -1;
@@ -947,13 +939,14 @@ static int read_relocs(struct elf *elf)
reloc->sec = sec;
reloc->idx = i;
- reloc->sym = find_symbol_by_index(elf, symndx);
+ reloc->sym = sym = find_symbol_by_index(elf, symndx);
if (!reloc->sym) {
WARN("can't find reloc entry symbol %d for %s",
symndx, sec->name);
return -1;
}
+ list_add_tail(&reloc->sym_reloc_entry, &sym->reloc_list);
list_add_tail(&reloc->list, &sec->reloc_list);
elf_hash_add(reloc, &reloc->hash, reloc_hash(reloc));
--- a/tools/objtool/include/objtool/elf.h
+++ b/tools/objtool/include/objtool/elf.h
@@ -62,6 +62,7 @@ struct symbol {
u8 fentry : 1;
u8 profiling_func : 1;
struct list_head pv_target;
+ struct list_head reloc_list;
};
struct reloc {
@@ -73,6 +74,7 @@ struct reloc {
};
struct section *sec;
struct symbol *sym;
+ struct list_head sym_reloc_entry;
unsigned long offset;
unsigned int type;
s64 addend;
next prev parent reply other threads:[~2022-11-02 21:47 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-10-28 19:40 [PATCH 0/5] x86/kallsyms: Fix the call-thunk kallsyms fail Peter Zijlstra
2022-10-28 19:40 ` [PATCH 1/5] kallsyms: Revert "Take callthunks into account" Peter Zijlstra
2022-10-28 19:40 ` [PATCH 2/5] objtool: Slice up elf_create_section_symbol() Peter Zijlstra
2022-11-02 9:20 ` [tip: x86/core] " tip-bot2 for Peter Zijlstra
2022-10-28 19:40 ` [PATCH 3/5] objtool: Avoid O(bloody terrible) behaviour -- an ode to libelf Peter Zijlstra
2022-11-02 9:20 ` [tip: x86/core] " tip-bot2 for Peter Zijlstra
2022-11-02 22:22 ` [PATCH 3/5] " Josh Poimboeuf
2022-11-03 8:46 ` Peter Zijlstra
2022-10-28 19:40 ` [PATCH 4/5] objtool: Add option to generate prefix symbols Peter Zijlstra
2022-11-02 9:20 ` [tip: x86/core] " tip-bot2 for Peter Zijlstra
2022-11-02 23:41 ` [PATCH 4/5] " Josh Poimboeuf
2022-10-28 19:40 ` [PATCH 5/5] x86: Add prefix symbols for function padding Peter Zijlstra
2022-11-02 9:20 ` [tip: x86/core] " tip-bot2 for Peter Zijlstra
2022-10-30 9:15 ` [PATCH 0/5] x86/kallsyms: Fix the call-thunk kallsyms fail Peter Zijlstra
2022-10-31 6:18 ` Yujie Liu
2022-11-02 21:46 ` Peter Zijlstra [this message]
2022-11-05 10:36 ` [tip: x86/core] objtool: Optimize elf_dirty_reloc_sym() tip-bot2 for Peter Zijlstra
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=Y2LlRA7x+8UsE1xf@hirez.programming.kicks-ass.net \
--to=peterz@infradead.org \
--cc=djwong@kernel.org \
--cc=joao.moreira@intel.com \
--cc=jpoimboe@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=samitolvanen@google.com \
--cc=tglx@linutronix.de \
--cc=x86@kernel.org \
--cc=yujie.liu@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®